Support infer_discrete for Predictive (#1086)

* support infer_discrete for Predictive

* revise docs

* use infer_discrete_temperature

* use temperature=1 by default

Author: fehiepsi

examples/annotation.py

@@ -42,10 +42,9 @@
 
 import numpyro
 from numpyro import handlers
-from numpyro.contrib.funsor import config_enumerate, infer_discrete
 from numpyro.contrib.indexing import Vindex
 import numpyro.distributions as dist
-from numpyro.infer import MCMC, NUTS
+from numpyro.infer import MCMC, NUTS, Predictive
 from numpyro.infer.reparam import LocScaleReparam
 
 
@@ -313,24 +312,9 @@ def main(args):
     mcmc.run(random.PRNGKey(0), *data)
     mcmc.print_summary()
 
-    def infer_discrete_model(rng_key, samples):
-        conditioned_model = handlers.condition(model, data=samples)
-        infer_discrete_model = infer_discrete(
-            config_enumerate(conditioned_model), rng_key=rng_key
-        )
-        with handlers.trace() as tr:
-            infer_discrete_model(*data)
-
-        return {
-            name: site["value"]
-            for name, site in tr.items()
-            if site["type"] == "sample" and site["infer"].get("enumerate") == "parallel"
-        }
-
     posterior_samples = mcmc.get_samples()
-    discrete_samples = vmap(infer_discrete_model)(
-        random.split(random.PRNGKey(1), args.num_samples), posterior_samples
-    )
+    predictive = Predictive(model, posterior_samples, infer_discrete=True)
+    discrete_samples = predictive(random.PRNGKey(1), *data)
 
     item_class = vmap(lambda x: jnp.bincount(x, length=4), in_axes=1)(
         discrete_samples["c"].squeeze(-1)

numpyro/contrib/funsor/discrete.py

@@ -118,8 +118,7 @@ def _sample_posterior(
             values = [v.reshape((-1,) + prototype_shape[1:]) for v in values]
             data[root_name] = jnp.concatenate(values)
 
-    with substitute(data=data):
-        return model(*args, **kwargs)
+    return data
 
 
 def infer_discrete(fn=None, first_available_dim=None, temperature=1, rng_key=None):
@@ -169,6 +168,12 @@ def viterbi_decoder(data, hidden_dim=10):
             temperature=temperature,
             rng_key=rng_key,
         )
-    return functools.partial(
-        _sample_posterior, fn, first_available_dim, temperature, rng_key
-    )
+
+    def wrap_fn(*args, **kwargs):
+        samples = _sample_posterior(
+            fn, first_available_dim, temperature, rng_key, *args, **kwargs
+        )
+        with substitute(data=samples):
+            return fn(*args, **kwargs)
+
+    return wrap_fn

numpyro/infer/util.py

@@ -11,12 +11,13 @@
 from jax import device_get, jacfwd, lax, random, value_and_grad
 from jax.flatten_util import ravel_pytree
 import jax.numpy as jnp
+from jax.tree_util import tree_map
 
 import numpyro
 from numpyro.distributions import constraints
 from numpyro.distributions.transforms import biject_to
 from numpyro.distributions.util import is_identically_one, sum_rightmost
-from numpyro.handlers import replay, seed, substitute, trace
+from numpyro.handlers import condition, replay, seed, substitute, trace
 from numpyro.infer.initialization import init_to_uniform, init_to_value
 from numpyro.util import not_jax_tracer, soft_vmap, while_loop
 
@@ -673,17 +674,47 @@ def _predictive(
     posterior_samples,
     batch_shape,
     return_sites=None,
+    infer_discrete=False,
     parallel=True,
     model_args=(),
     model_kwargs={},
 ):
-    model = numpyro.handlers.mask(model, mask=False)
+    masked_model = numpyro.handlers.mask(model, mask=False)
+    if infer_discrete:
+        # inspect the model to get some structure
+        rng_key, subkey = random.split(rng_key)
+        batch_ndim = len(batch_shape)
+        prototype_sample = tree_map(
+            lambda x: jnp.reshape(x, (-1,) + jnp.shape(x)[batch_ndim:])[0],
+            posterior_samples,
+        )
+        prototype_trace = trace(
+            seed(substitute(masked_model, prototype_sample), subkey)
+        ).get_trace(*model_args, **model_kwargs)
+        first_available_dim = -_guess_max_plate_nesting(prototype_trace) - 1
 
     def single_prediction(val):
         rng_key, samples = val
-        model_trace = trace(seed(substitute(model, samples), rng_key)).get_trace(
-            *model_args, **model_kwargs
-        )
+        if infer_discrete:
+            from numpyro.contrib.funsor import config_enumerate
+            from numpyro.contrib.funsor.discrete import _sample_posterior
+
+            model_trace = prototype_trace
+            temperature = 1
+            pred_samples = _sample_posterior(
+                config_enumerate(condition(model, samples)),
+                first_available_dim,
+                temperature,
+                rng_key,
+                *model_args,
+                **model_kwargs,
+            )
+        else:
+            model_trace = trace(
+                seed(substitute(masked_model, samples), rng_key)
+            ).get_trace(*model_args, **model_kwargs)
+            pred_samples = {name: site["value"] for name, site in model_trace.items()}
+
         if return_sites is not None:
             if return_sites == "":
                 sites = {
@@ -698,9 +729,7 @@ def single_prediction(val):
                 if (site["type"] == "sample" and k not in samples)
                 or (site["type"] == "deterministic")
             }
-        return {
-            name: site["value"] for name, site in model_trace.items() if name in sites
-        }
+        return {name: value for name, value in pred_samples.items() if name in sites}
 
     num_samples = int(np.prod(batch_shape))
     if num_samples > 1:
@@ -729,6 +758,12 @@ class Predictive(object):
     :param int num_samples: number of samples
     :param list return_sites: sites to return; by default only sample sites not present
         in `posterior_samples` are returned.
+    :param bool infer_discrete: whether or not to sample discrete sites from the
+        posterior, conditioned on observations and other latent values in
+        ``posterior_samples``. Under the hood, those sites will be marked with
+        ``site["infer"]["enumerate"] = "parallel"``. See how `infer_discrete` works at
+        the `Pyro enumeration tutorial <https://pyro.ai/examples/enumeration.html>`_.
+        Note that this requires ``funsor`` installation.
     :param bool parallel: whether to predict in parallel using JAX vectorized map :func:`jax.vmap`.
         Defaults to False.
     :param batch_ndims: the number of batch dimensions in posterior samples. Some usages:
@@ -749,10 +784,12 @@ def __init__(
         self,
         model,
         posterior_samples=None,
+        *,
         guide=None,
         params=None,
         num_samples=None,
         return_sites=None,
+        infer_discrete=False,
         parallel=False,
         batch_ndims=1,
     ):
@@ -801,6 +838,7 @@ def __init__(
         self.num_samples = num_samples
         self.guide = guide
         self.params = {} if params is None else params
+        self.infer_discrete = infer_discrete
         self.return_sites = return_sites
         self.parallel = parallel
         self.batch_ndims = batch_ndims
@@ -838,6 +876,7 @@ def __call__(self, rng_key, *args, **kwargs):
             posterior_samples,
             self._batch_shape,
             return_sites=self.return_sites,
+            infer_discrete=self.infer_discrete,
             parallel=self.parallel,
             model_args=args,
             model_kwargs=kwargs,